Lab Week 5: Time series and a map!

Attach packages

library(tidyverse)
library(janitor)
library(lubridate)
library(here)
library(paletteer)
library(tsibble)
library(fable)
library(fabletools)
library(feasts)
library(forecast)
library(sf)
library(tmap)
library(mapview)

Monthly US energy consumption (renewables)

us_renew <- read_csv(here("data","renewables_cons_prod.csv")) %>% 
  clean_names()

• Make description all lowercase
• Only keep observations where “description” variable contains “consumpiton”
• Remove any observations where “description” variable contains “total”

renew_clean <- us_renew %>% 
  mutate(description = str_to_lower(description)) %>%  #use mutate to over-right
  filter(str_detect(description, pattern = "consumption")) %>% #now look for a pattern "consumption" and if isn't there get rid of that row (string detect is a true false logical function (if tru then it stays, if false then it is taken out))
  filter(!str_detect(description, pattern = "total"))#because we now want to exclude, or look for false use !

Convert ‘yyymm’ column to a date AND then help it know that it’s a time frame

#Now we want to work with time component!
#Lubridate is great and makes nice assumptions but you should always check (like month 13! It just didn't make dates for those because that's rediculous)

renew_date <- renew_clean %>% 
  mutate(yr_mo_day = lubridate::parse_date_time(yyyymm, "ym")) %>%  #give it the structure of the colum and formae
  mutate(month_sep = yearmonth(yr_mo_day)) %>% #now converting to SIBBLE format because FEAST AND FORMAT wants it with SIBBLE, month_sep is stored as a DATE and YEARMON which is great for time series stuff later
  mutate(value = as.numeric(value)) %>%  #value was read in as a character
  drop_na(month_sep, value) #where month_sep OR value has an NA, get rid of that row

#Make a version where I have the month & year in separate colums: 
renew_parsed <- renew_date %>% 
  mutate(month = month(yr_mo_day, label = TRUE)) %>% #label = true means that the new month is a class of ordered factor that has the month abbreviations
  mutate(year = year(yr_mo_day))

Look at it:

renew_gg <- ggplot(data = renew_date, aes(x = month_sep, 
                                          y = value,
                                          group = description)) +
  geom_line(aes(color = description))


renew_gg

Updating colors with paleteer palettes: SO FREAKING COOL

renew_gg + 
  scale_color_paletteer_d("ggsci::uniform_startrek")#chood d because we are choosing from DISCRETE functions

Coerce renew_parsed to a tsibble

renew_ts <- as_tsibble(renew_parsed, key = description, index = month_sep)#index is the sibble compatable time variable (in this case month_sep)

Let’s look at our ts data in a couple different ways:

renew_ts %>% autoplot(value) #because we told it above what to use, this does all the other work

renew_ts %>% gg_subseries(value) #see the different sources broken up by month across all years, graphing exploration super fast!

#renew_ts %>% gg_season(value) #seasonally, but this breaks a lot so let's make this another way:

season_graph <- ggplot(data = renew_parsed, aes(x = month, y = value, group = year)) +
  geom_line(aes(color = year)) +
  facet_wrap(~description,
             ncol = 1,
             scales = "free",
             strip.position = "right")

Just look at hydroelectric energy consumption

hydro_ts <- renew_ts %>% 
  filter(description == "hydroelectric power consumption")

hydro_ts %>% autoplot(value)

hydro_ts %>% gg_subseries(value) #blue lines are monthly values! 

#hydro_ts %>%  gg_season(value)

ggplot(hydro_ts, aes(x = month, y = value, group = year)) +
  geom_line(aes(color = year))

What if I want quarterly average consumption for hydro?

hydro_quarterly <- hydro_ts %>% 
  index_by(year_qu = ~(yearquarter(.))) %>% #group as a function of yearquarter based on (.) the groups that exist 
  summarize(ave_consumption = mean(value))

head(hydro_quarterly)

## # A tsibble: 6 x 2 [1Q]
##   year_qu ave_consumption
##     <qtr>           <dbl>
## 1 1973 Q1            261.
## 2 1973 Q2            255.
## 3 1973 Q3            212.
## 4 1973 Q4            225.
## 5 1974 Q1            292.
## 6 1974 Q2            290.

Decompose that hydro_ts

dcmp <- hydro_ts %>% 
  model(STL(value ~ season(window = 5))) #model value as a function of season

components(dcmp) %>% autoplot() #+ #top is acutaly data, then moving average tren, then seasonal extracted, then fourth graph is anything is left over 

  #scale_y_continuous(limits = c(150,350)) #you can force scales like you would in gg plot (not always helpful)

hist(components(dcmp)$remainder)

Now look at the ACF:

hydro_ts %>% 
  ACF(value) %>% 
  autoplot()

#x axis lab[1M], means that it is showing month by month and that every 12 months are most coorelated 

DANGER DANGER DANGER (don’t use this unless you know your shit)

#hydro_model <- hydro_ts %>% 
 # model(
   # ARIMA(value),
    #ETS(value)
 # ) %>% 
 # fabletools::forecast(h = "4 years") #how long into the future since your last point you want to forcast

#hydro_model %>% autoplot(filter(hydro_ts, year(month_sep) > 2010))#want to tack on existing data and can choose time range

Make a world map!

customizing use tmap, quick and dirty look use mapview

#world <- read_sf(dsn = here("data", "TM_WORLD_BORDERS_SIMPL-0.3-1"),
                 #layer = "TM_WORLD_BORDERS_SIMPL-0.3")

#mapview(world)